The problem for adding a +/-15V supply in an existing amplifier with main voltage rails of +/-50V to +/-60V, for instance; is using 78XX/79XX voltage regulators, or even LM317T/LM337T. The former has a maximum input voltage of +/-35V, and the latter +/-40V.

Even using high voltage regulators (if you can still get them) will dissipate quite a fair amount of heat dropping the voltage down to 15V. Wasted power!

We could also use a simple Zener shunt regulator. The problem with this is, with a 1W Zener, current draw is only good up to about 40mA before the voltage at the output starts to decrease. The other issue is Zener's generate noise.

So, let's look at a basic solution.

Fig. 1: basic series-pass voltage regulator

The key component is the Zener diode D1, which is a 15V zener. Resistor R1 supplies current from the +50V rail into the Zener diode. The Zener then clamps the transistor base voltage to approximately:

Vb = 15V

Capacitor C2 filters noise and ripple on this Zener reference voltage, making it cleaner and more stable.

Transistor Q1 is connected as an emitter follower, also called a series-pass transistor. In an emitter follower, the emitter voltage follows the base voltage minus the transistor's base-emitter voltage drop, which is typically about 0.6 to 0.7V.

So the output voltage becomes approximately:

Vout = 15V − 0.65V = 14.35V

which matches the labeled output voltage of about +14.55V.

The transistor serves several important purposes. First, it allows the circuit to supply much more output current than the zener diode alone could provide. The zener only controls the transistor base current, while the transistor supplies the larger load current from collector to emitter. Second, the transistor isolates the output from variations and noise on the incoming +50V rail.

Capacitor C1 filters the incoming supply and reduces ripple or noise from the main power source. Capacitor C3 filters the output rail and improves stability and transient response.

This type of circuit is often called a Zener-transistor regulator or emitter-follower regulator. It is simple, inexpensive, and commonly used in analog audio equipment where a reasonably stable supply is needed but ultra-precise regulation is not critical.

One important limitation is that the output voltage depends on the transistor’s base-emitter voltage, which changes slightly with temperature and current. So the regulation is not as accurate as an integrated regulator like a 7815, but for many analog circuits it works very well.

The one main issue is, with an output current of 15mA, there will be a dissipation of 525mW in Q1 (as it's dropping the supply to 35V) - so a large heatsink would be required, as it will get hot. Another issue, apart from heat, is the more current that is drawn on the output, the more the voltage begins to drop. To power a preamplifier, for instance, this shouldn't be that much of an issue. However, the Zener is going provide a lot of noise in the circuit, which the transistor will happily amplify.

However, let's look at a solution to the voltage drop problem.

Fig. 2: series-pass regulator with error amplifier

This circuit is a more advanced transistor linear regulator than the previous one. It converts an unregulated +50V supply into a much more stable +15V output using a feedback-controlled series-pass regulator.

The main improvement over the simpler Zener emitter-follower regulator is the addition of Q2, which acts as an error amplifier. Instead of the output voltage depending mostly on the transistor’s base-emitter voltage, this circuit actively compares the output voltage against a Zener reference and continuously corrects any error.

The output voltage is controlled by adjusting the base current of Q1 and Zener diode D1 provides a stable voltage reference of approximately 7.5V. Resistor R4 supplies current from the output rail into the Zener diode so the it remains in regulation.

Transistor Q2 is the key control element. Its emitter is tied to the Zener reference voltage, so the emitter sits near 7.5V. The base of Q2 monitors a divided-down version of the output voltage using the resistor divider made from R2 and R3.

The divider voltage is:

Vb = Vout * R3 / R2 + R3

Using the resistor values R2 of 470 ohm and R3 of 560 ohm, the divider ratio becomes:

470 + 560 / 560 = 0.544

So when the output is about 15V:

15V * 0.544 = 8.16V

Since Q2 is an NPN transistor, it conducts when:

Vb = Ve + 0.65V

With the emitter at about 7.5V, conduction begins when the base reaches approximately:

7.5V + 0.65V = 8.15V

which matches the divider voltage at roughly a 15V output. This is the regulation point.

Let's now discuss how the feedback action works.

If the output voltage rises above 15V:
- the divider voltage at the base of Q2 rises
- Q2 conducts harder
- Q2 pulls current away from the base of Q1
- Q1 conducts less
- the output voltage falls back down

If the output voltage drops:
- the divider voltage decreases
- Q2 conducts less
- more current flows through R1 into the base of Q1
- Q1 conducts harder
- the output voltage rises again

So Q2 continuously adjusts the base drive of Q1 using negative feedback to stabilize the output voltage.

However, again at 15mA of current draw on the output, Q1 will be dissipating around 600mW - so, heat is still a big factor here, even at low currents. This would again mean a large heatsink would be required. The Zener diode will still be injecting quite a bit of noise in to the circuit which Q1 will amplify.

There is another issue with the circuits shown in Fig. 1 and 2, apart from the dissipation and noise issues. There's no short-circuit protection, so if the output was shorted to ground - the series-pass transistors (and possibly the Zener) will be destroyed.

Now we'll look at a solution which combines a simple series-pass initial “pre-regulator” and the 7815/7915 series linear regulators and hopes to address all the above issues.

Fig. 3: The final low-noise approach of a series-pass +/-15V regulator to power a preamplifier

The circuit works in two stages. First, the voltage is reduced from +/-50V down to about +/-26.5V using series-pass pre-regulators. Then the 7815 and 7915 linear regulators further regulate this down to precise +/-15V rails, and also help to reduce noise.

The reason the pre-regulator transistors Q1 and Q2 are important is because dropping voltage directly from +/-50V to +/-15V inside the 7815 and 7915 would create excessive power dissipation and heat (if not destroy them, as the maximum input voltage is +/-35V). For example, without pre-regulation the regulators would need to drop:

50V − 15V = 35V

which would generate substantial heat even at modest current.

With the pre-regulator stage, the regulators only drop about:

26.5V − 15V = 11.5V

greatly reducing regulator dissipation and improving reliability.

The two resistors R1/R2 should be 1% metal film type to help reduce noise, and as they have (at +/-50V) a voltage drop of around 23V - the current through them is going to be around 5mA (0.115W). 1/4W resistors would be “OK”, but 1/2W would be better. The dissipation through the transistors will be considerably less also as the voltage drop it's dealing with (23V base-collector) is less than in Fig. 1 or 2.

A heatsink would still be required for the transistors, and the 7815/7915 may not need heatsinking at low currents (under 100mA); but it's not a bad idea to mount them to the same heatsink as the transistors. This will allow the IC regulator's internal thermal overload shutdown circuitry to monitor the series-pass elements.

With the use of the 7815/7195 linear regulators as the “final” outputs, this also takes advantage of their internal short-circuit protection circuitry. So, in the event of a short to ground, for instance, the series-pass transistors in the pre-regulators are not going to be destroyed. However, it's never really a good idea to intentionally short the output of a power supply, even for testing.

Overall, this circuit is a clean, low-noise dual tracking supply designed specifically for powering sensitive low-level audio circuitry from a much higher-voltage amplifier supply rail.

Even though the above circuit is a valid solution, generally speaking this is why a lot of high-end amplifiers use separate transformers. One for the high-current/voltage power amplifiers, and a small one (say a 15-0-15V AC) for the lower current preamplifier(s) - with regulation. That's not to say the circuit doesn't have merit, or it isn't a solution if you don't want to add a second transformer.

All circuits shown here were simulated in TINA-TI and not built or tested.